time-resolved biological and perturbation chemical crystallography… · 2019-11-27 ·...

Invited Paper

Time-resolved biological and perturbation chemical crystallography:Laue and monochromatic developments

G. Bradbrook, A. Deacon, J. Habash, J.R. Helliwell*, M. Helliwell, Y.P. Nieh,

E.H. Snell, & S. Trapani

Department of Chemistry, University of ManchesterManchester M13 9PL, U.K.

A.W. Thompson & J.W. Campbell

Daresbury Laboratory, Daresbury, Warrington, Cheshire, WA4 4AD, UK

N.M. Allinson & K. Moon

Department of Electronics, University of York, Heslington, York, YOl 5DD, UK

T. Ursby & M. Wulif

ESRF, BP220, Grenoble Cedex, France

*To whom correspondence should be addressed

+Present address: EMBL/ESRF, Avenue des Martyrs, Grenoble Cedex, France

ABSTRACT

Time-resolved macromolecular X-ray crystallography is a new capability for structural analysis driven bycontinuing improvements in synchrotron X-ray sources, optics and detectors (image plates and CCDs). Proteincrystal Laue data (stationary crystal and polychromatic X-rays) were recorded at SRS Daresbury station 9.5 andESRF Grenoble beamline 3, and processed with the Daresbury Laue software package. The Laue method allowsexposure times set by the synchrotron electron bunch width, eg. 50 picoseconds. The instruments and methodsdevelopments widen opportunities for perturbation chemical crystallography studies too. A temperature depen-dent phase transition of a liquid crystal nickel-octahexylphthalocyanine is studied with a rapid readout CCDdetector. Structure solution by molecular replacement methods with Laue data is reported for orthorhombiclysozyme. By use of tetragonal lysozyrne as a test case it is shown that with fine angular intervals, wide totalangular coverage of Laue exposures and the deconvolution of multiples, good connectivity of electron densitymaps can be realised. The monochromatic rotating crystal method offers possibilities of extremely fast rotationswhich allow a complete data set to be recorded onto a single image -Large-angle oscillation technique (LOT).The processed LOT data looks promising. LOT electron density maps are presented. We are now exploring theuse of 3600 rotations onto a single image. However, since the time required to cover one revolution of the crystal

160 / SPIE Vol. 2521 0-8194-1880-3/951$6.OO

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 02/17/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

is likely to be greater than 1 millisecond, and because the monochromatic synchrotron radiation beam intensitysets its own limits to the shortest exposure time, the Lane method will retain its role for the fastest exposures incrystallography. Laue techniques open up quicker data collection opportunities in neutron protein crystallography.

Keywords: synchrotron radiation, CCD detector, liquid crystal transition, time-slicing, Laue method, LOTmethod, proteins, electron density, molecular replacement, lysozyme, neutrons. concanavalin A.

1 INTRODUCTION

Synchrotron radiation is now used extensively in macromolecular crystallography.' The first use of syn-chrotron radiation as a source for X-ray diffraction was directed at time-resolved biological fibre diffraction forthe monitoring of the dynamics of muscle contraction.2 Early considerations of the use of synchrotron radiationto measure data from crystalline samples included dynamical studies.3 Data collection from protein crystals ingeneral faces the major problems of weakness of reflections, a voluminous reciprocal space to be surveyed andradiation damage.3 Time-resolved macromolecular crystallography4 imposes yet further, major, demands sincethe quality of the determined protein structure must remain adequate whilst the single crystal data are measuredrapidly and repeatedly, i.e. at specific time points.5

The fastest method of data collection uses a stationary crystal and a white X-ray beam (Laue method)6; itallows exposure times set by the synchrotron bunch width to measure a sizable region of diffraction (reciprocal)space but a full coverage is not realised in a single exposure. Hence a distinction is usefully made between X-rayexposure time and elapsed data collection time.5 Better completeness is obtained, especially at low resolution,by monochromatic rotating crystal methods. By harnessing the collimation property of synchrotron radiationvery large rotations can be covered in a single diffraction image; such a large-angle oscillation technique (LOT)allows the whole lot of data to be recorded onto a single image thus avoiding any time overheads of detectorreadout time.7 The simultaneous translation of the detector and rotation of the crystal, over a more limitedangular range, the Weissenberg method, as employed at synchrotron radiation sources8 is also a monochromaticmethod advocated for fast data collection.9 The Weissenberg method avoids overlapping of reflections but coversa restricted range of reciprocal space. LOT (section 4) offers a full coverage of reciprocal space but overlap ofreflections is a problem at high resolution. The Laue method can achieve a high completeness at high resolutionbut the completeness at lower resolution is worse. A combination of Laue and LOT, with their complementarycharacteristics is therefore of interest. However, the Laue method has the pre-eminent role of allowing the fastestexposure times possible because all of the spectral emission from the synchrotron is condensed onto the crystalsimultaneously in contrast to monochromatic rotating crystal methods where a tiny proportion of the synchrotronradiation emission is used (although this can be made more optimal by the use of an undulator to compress thewhite spectrum into specific, interference peaks in the emitted spectrum'). According to the type of structuralchange to be studied (reversible or irreversible) and its time constant the different tools, just referred to, oftime-resolved simple crystal data collection can be applied. Time-resolutions of 50 picoseconds are accessible, atthe ESR.F 'Laue beamline' (BL3) where enough X-rays are focused onto a crystal to achieve a Laue photographfrom one electron bunch circulating the ring.

The historical objection to the Laue method, that most of the Laue diffraction spots would be multiple(whereby Laue spots would comprise several reflections from interplanar spacings of d, , etc. satisified by thosewavelengths in the white spectrum .X, , etc.)6"° has been analysed in detail'1; the multiple orders problem isevidently not as serious as was initially believed since the proportion of reflections occurring alone in Laue spotsnever falls below 72.8%.h1 It is the remainder, however, that contribute to an under sampling of reflections at lowresolution (large d spacing) as well as the shape of the volume of reciprocal space sampled by the Laue method.Improvements in detectors and computing, allow a much finer angular sampling of reciprocal space than hitherto.This will be dealt with in detail in section 3.4 (also section 2.2). The precise quantitation of Laue diffraction

SPIE Vol. 2521 / 161


intensity data has required the derivation and application of a wavelength-dependent correction (A curve), to takeinto account a variety of source, instrument and sample induced variations in intensity at wavelengths in the broadbandpass. Strategies'2 have been put forward and methods'3 developed for singles and deconvolution of multiplesin a Laue data set. These have been adequate to reveal a single water molecule in a protein of 30,000 molecularweight'4 and an accurate refinement of CO-myoglobin (including a partially occupied water molecule).'5 Furtherevidence that accurate Laue data and structural results are obtainable are provided in this paper (section 3).

2 INSTRUMENTATION

2.1 Low emittance sources and beamline designs for crystallography: Station 9.5of the Daresbury SRS

Most of the analyses (Laue and LOT) presented here have been carried out with synchrotron data collectedat station 9.5. This station design'6 takes advantage of the improvement of the emittance of the SRS (theso-called high brightness lattice) made in the mid 1980's whereby the point focusing function of the bearnlineoptics is completely provided by the grazing incidence, doubly curved, mirror. Hence a point focused white beam(O.5A < A < 2.OA) can be obtained of extremely high intensity opening up millisecond Laue crystallography (seefig.2 of ref.'7). Moreover for monochromatic data collection rapid tunability is achieved via a double crystal,water cooled, channel cut monochromator and so multi-wavelength anomalous scattering data collection can bemade'8 as well as high resolution data collection at short wavelengths (e.g. O.7A at SRS) much easier than onthe other instruments at the SRS, 7.2's and 9.6.20

The station 9.5 beatnline design concept lends itself to new, yet lower, emittance sources, like the ESRF(section 2.2), where the point focusing of the mirror allows the naturally small source size area to be even bettermatched to typical collimator(crystal) sizes (O.2 mm). Very recent hardware improvements in the 9.5 hutchhave been made consisting of a 3 axis goniometer2' for protein crystal alignment which are certainly welcomeallowing particular crystal sample settings in monochromatic22 and Laue data collection23 to be explored. Thework reported in this paper (sections 3 and 4) is based, however, on a low cost, single axis, crystal orientationdevice.

In summary, the overall trend is for source emittance (source size x source divergence) to decrease. This hasfitted the needs of protein crystallography extremely well as the sample acceptance (sample size x mosaicity) hasthus become better matched by the source characteristics available.

2.2 Extremely fast exposure times in biological crystallography: The ESRF comesonline (Beamline 3 and CCD detector)

The ESRF offers unprecedented brilliance, brightness and flux in the X-ray region for macromolecular crys-tallography,24 extending up to high photon energies, compared with other current X-ray sources. Moreover thecollimation can essentially reach a plane wave with a reasonable beam intensity at the sample. Combining suchthird generation synchrotron sources with detector developments such as the charge-coupled device (CCD) hasand will continue to result in improved detector time response and quality of data available.

We have conducted experiments on BL325 of the ESRF with a CCD26 detector probing the data quality thatcan be obtained. For example, nine Laue images from tetragonal lysozyme were collected at 10 intervals, 20millisecond exposure time, for calibration and spatial uniformity response checks. The data were processed usingthe Daresbury Laue suite13 to 2A resolution in the wavelength range 0.28 to 2.5A. For wavelength normalisation a

162 ISPIE Vol. 2521


range of 0.43 to O.90A was used. The normalisation curve is shown in figure la, the peak at O.55A is at a very muchshorter wavelength than data collected at previous synchrotron sources such as the SRS wiggler.'3 The Pergeis 8.8% on I. By comparing the wavelength-normalised data from one image with a reference monochromatic

(a)

Figure 1: ESRF BL3 with CCD Laue data from a tetragonal lysozyme test crystal (a) Wavelength normalisationcurve, wavelength is given in A. (b) Contour plot of the spatial response derived from LAIJESCALE of the CCDdetector; contours are in 5% response levels down from 100% at the centre of the detector.

(diffractometer) data set using the Daresbury program LAUESCALE it is possible to calculate the detector non-uniformity of response. One such contour plot is shown, figure ib, which is applied to the raw image and thedata reprocessed (a sort of post-refinement technique).

2.3 Rapid readout CCD detectors and perturbation chemical crystallography: Spacegroup transition of nickel-octahexylphthalocyanine as a function of tempera-ture

There exists considerable promise for the use of CCD imagers in the fast recording of diffraction patterns. Thesedevices complement the use of micro-spectrophotometric monitoring of reactions in crystals where a chromophoreis present but uniquely give insight into the crystal diffraction quality. When a chromopliore is not presentsuch time-slicing could be the only guide to crystal structure monitoring as a function of time. As part of thedevelopment of such devices we have described the first tests of very short (80 millisecond) protein crystal Lauediffraction oses27 Subsequently tests of intensity reproducibility showed merging R-factors on I of 3%•28 Atime-resolved study of radiation damage to a protein crystal was then made involving the repeated measurementof a part of the Laue pattern over a period of 4.2 seconds and showed the steady increase in crystal mosaic spreadwith exposure.5 These developments offer application too in perturbation chemical crystallography. A review ofthis topic can be found in ref.29

Here we report the use of a CCD detector30 to determine the transition temperature from one crystal spacegroup to another, of nickel octahexylphthalocyanine. This becomes a therinotropic liquid crystal at 140°C.Between room temperature and liquid nitrogen temperature the crystal undergoes another transition. By startingat lOOK and raising the temperature stepwise, the CCD diffraction pattern was recorded at each temperature onstation 7.2 of the SRS19 with a monochromatic beam. The CCD was coupled to a 2.6:1 fibre optic taper with astructured CsI(Tl) scintillator grown directly onto the front. The image consists of 576 x 384 pixels with eachpixel being 58.5 m square giving a total aperture of 33.7 x 22.5 mm. The system is optimised for fast readout,

SPIE Vol. 2521 / 163

0.55 0.60 0.65 0.70 0.75 0.60 0.65WaveI.nh

(b)


ct1OO milliseconds at 12 bit dynamic range. A disadvantage of the test made with the monochromatic beam wasthe rather long exposure times required which led to a noise build up. This would not be a problem in generalwith a polych.romatic beam, or a stronger monochromatic beam. Nevertheless the experiment performed allowedthe determination of the transition temperature within only a few hours of beamtime. A disappearance of spotsdue to h + k = 2n -F 1 occurred between 195 to 200K representing the change between the low temperatureprimitive crystal lattice and the higher temperature C-centred crystal lattice (figure 2). The structural details ofthis transition will be described elsewhere.

I r *; &jflrt "I ' s4** \ _I '- c9;_ ra t .- tljc- M: 14'; j:; ;. :-'j ;—tt' — Iz& r : — ,- .&4'z:•t&ttA:cfir_ C::Jcr ;+ .-**L' , t t

C:t$ta *::c:j*sf%rraa: :%tc P a*Strc- 't, .. . -t.. • srtc.tFe r tS -- \ #__4i A- — Z

A4 _ r _ . . ,c;a; _ :c !t; IL ;_ '

Figure 2: Two of the CCD images showing a disappearance of spots due to h + k = 2n + 1 occurring between195K (left) to 200K (right) representing the change between the low temperature primitive crystal lattice andthe higher temperature C-centred crystal lattice

3 LAUE PROTEIN CRYSTAL STRUCTURE ANALYSES: NEWDEVELOPMENTS AND RECENT CHALLENGES

Recent challenges to the Laue method for structural studies of proteins have included the impact of thedistribution. of those reflections tied up in 'multiple' Laue spots," particularly with respect to fragmentation(connectivity) of certain electron density maps (e.g. 2Fo-Fc), and also identifying atoms with high B-factors ina structure. Moreover, our wavelength normalisation' has been criticised32 in terms of the uncertainty of thescale factors, in spite of the concordance of scale factors shown in our ('Daresbury' package) broad wavelengthband protein crystal Laue data and processing'3 and the structural results obtained.'4"5

Of considerable interest to us have been the new experimental opportunities for Laue data collection affordedby automatic readout detectors, and of enhanced sensitivity, such as on-line image plates and CCDs. The enhancedsensitivity allows fairly small protein crystals to be used for Laue data collection (section 3.2). These detectorsalso circumvent the labour of Laue photography. Hence a much finer angular interval between Laue exposures canbe obtained within the required coverage of a given crystal symmetry. This will allow a much better completenessat low resolution to be more readily achieved (examples at different angular intervals are given in section 3.4below).

164 ISPIE VoL 2521


3.1 Obtaining higher resolution Laue data: 2.OA versus 2.5A tetragonal lysozyme

Obtaining high resolution Laue data with a broad bandpass beam is practical as evidenced in our previousstudy of carbonic anhydrase14 (2.2A high quality data). Likewise, for native concanavalin A, broad bandpassLaue data yielded a resolution limit of 2.4 A, but restricting the bandpass (using an Al filter) however, reducedthe background allowing a prolonged exposure time yielding data to 1.95A.7 An automatic control of )maz isclearly important for bandpass limitation and for this reason we have developed a transzthssion mirror basedon thin mylar films.33'34 We now report on similar results for chicken egg white lysozyme. By restricting thebandpass (0.2 mm Al filter) and using longer exposure times (750 millisececonds versus 250 milliseconds), forsimilar machine ring currents, the resolution limit is enhanced from 2.5A to 2.OA.

A tetragonal lysozyme crystal (a=b=79.1A, c=37.9A, P43212) was used to collect Laue data on station 9•516of the SRS. The SEtS operated at 2.0 GeV and 196 mA with the wiggler at 5T. The collimator diameter of 0.2mm was smaller than the crystal in each dimension. Data were collected on an image plate detector (MARresearch) of 90 mm radius placed 204 mm from the crystal. Five images were collected in 12° degree intervalscovering 0-48g. Laue images were processed using the Daresbury Laue suite of programs13 to obtain singles data,deconvoluted multiples data and finally a combined set of singles and deconvoluted multiples. Deconvolutionused the wavelength normalisation curve method.35 At the normalisation stage reflections stimulated within thewavelength range 0.52-1.60A were used. The combined singles and deconvoluted multiples data set contained6342 reflections with completeness dmjn 00 of 74.1%, d —2d of 77.9% and 2d,, —oo of 50.5% with anP1merge Of 8.8%. For comparison the singles data had a completeness d,, oo of 67.7%, d, — of 72.5%and 2d,., — 00 of 37.4%.

The original Protein Data Bank (PDB filename 6LYZ) coordinates36 were refined with the X-PLOR program.37Each cycle of refinement included 40 cycles of positional refinement followed by 20 cycles of temperature factorrefinement. Map calculations were carried out with the CCP4 program suite38 and maps were inspected with theprogram FRODO39 and later O.° Refinement converged with an R-factor of 21.0%, geometric rms deviationsof o.oiiA for bond lengths, 2.0° for bond angles and 22.6° for torsion angles. The average temperature factorsfor the main and side chain atoms were 13.3A2 and 17.7A2 respectively. There were 80 water molecules withan average temperature factor of 25A2. Inspection of the Ramachandran plot4' showed 100% of all nonglycineand non-proline residues lying in the most favoured and additional allowed regions. For comparison the singletLaue data set yielded a final Rfactor of 19.3% in model refinement for 5792 reflections with 63 common watermolecules in good density and chemically sensible positions.

The study showed in general that the combined data set gave improved electron density connectivity for themain and side chains. For TRP, PHE, TYR, HIS and PRO residues, the combined data set gave pronouncedimprovements in the electron density but did not enhance the number of water molecules (80 versus 77). Thebenefits in reaching higher resolution are evident in the increase in the number of reflections over earlier studies42at 2.5A (2573 single, 2963 combined) for refinement of the model.

3.2 Small protein crystals: Synchrotron Laue study of ortliorhombic lysozyme in-.cluding molecular replacement

Orthorhombic chicken egg white lysozyme crystals (obtained unexpectedly from the crystallization conditionsfor tetragonal lysozyme43) have been studied using synchrotron Laue diffraction. Laue data to 3A were collectedon station 9.5 of the SRS'6 on a relatively small crystal ( 0.08 x 0.10 x 0.50 mm) using an image plate (MARresearch). The SRS was operating at 2.0 GeV and 262 mA, a 0.2 mm thick Al foil was used to attenuate thelong wavelength (radiation damaging) parts of the beam giving a band pass of 0.4 < A <2.OA. The collimatordiameter was 0.2 mm. A total of 15 exposures were recorded over a range of 140° in 10° intervals with a 400millisecond exposure time per image. Unit cell dimensions for P212121 orthorhombic lysozyme44 (a=59.4A,

SPIE Vol. 2521 / 165


b=68.7A, c=30.8A) were used in the determination of crystal orientation with LAUEGEN.45 An example of oneimage obtained is shown in figure 3a. The data were processed using the Daresbury Laue software suite13 andreduced using the CCP4 suite of programs38 giving an overall R-factor on I of 7.2%, 1843 unique reflections andaverage multiplicity of 5.8. The completeness ranges are — cc of 88.4%, and 2dmjn — cc of 37.3%.

(b)

Figure 3: (a) Laue diffraction pattern from a small crystal of orthorhombic lysozyme (b) The resulting structureshowing the accessibility of saccharide binding sites D, E and F (indicated by in the centre) via the solventchannels for orthorhombic lysozyme.

These data were used to provide an initial molecular model using the MERLOT molecular replacement pack-age.46 The starting model was that of 2.OA resolution tetragonal lysozyme deposited at the PDB (6LYZ).36 Therotation and translation results obtained were as follows:

Euler rotation angles(°) 108.00 67.00 34.00Translation (as a fraction of unit cell edge) 0.466 0.367 O.OOfl

After rigid body refinement of the molecular replacement solution the R-factor on F was 40% indicating thatan essentially correct solution had been found. Graphics images of this crystal form of the enzyme clearly showthat the saccharide binding sites D, E and F are accessible via solvent channels (figure 3b), which is in agreementwith the earlier, MIR monochromatic results to 6A.44

Data have also been collected to a higher resolution (2A) at the NSLS and ESRF on sugar soaked crystalswith a view to undertaking the detailed refinement of the protein in this crystal form and hopefully complexedwith sugar. On a technical note it is impressive that crystals of this size could be used and is testimony to theimproved sensitivity of modern detectors (such as image plates and CCDs) over photographic film.

166 ISPIE Vol. 2521

(a)


3.3 Improved Laue software procedures: Reprocessing and re-refinement of cobaltconcanavaJin A using Laue image plate data

Using an improved version of the Daresbury Laue software suite including LAUEGEN45 we have repeated aprevious study33 on Co concanavalin A. A larger number of reflections has then been made available.

Laue data from a single crystal of Co concanavalin A (a=88.7A, b=86.5A, c=62.5A, 1222) were collected33 onstation 9.5 of the SRS'6 using a relatively narrow bandpass (O.5A< A < O.9A) to a resolution of 2.OA. A total offifteen images were collected covering an angular range of 122° and processed with the Daresbury Lane suite ofprograms.13

The singles data were reduced using the CCP438 suite of programs giving completeness of d — oo, 80.8%,dmin 2dmin of 84.1% and 2d — oo of 59.0% with an R-factor on I of 7.9%. The previous study gave acompleteness of d — oo, 76.5%, d — 2d of 80.5% and 2d — oo of 51.6% with an Rfactor on I of7.7%. In terms of the number of independent reflections, with the improved software we obtained 13436 uniquereflections compared with the 12723 reflections from the previous study.

A starting set of coordinates from a monochromatic 2A study of a Cd substituted concanavalin A, also inspace group I222 has been used for the structure refinement.

The water molecules were deleted from the start-ing coordinate set, the positional and B-factor re-finement has been carried out using X-PLOR.,37 fol-lowed by difference Fourier map calculation38 andmodel building using o.° The procedure has beenrepeated until convergence. During the model build-ing residues 151 and 155 have been changed fromGlu to Asp and Arg to Glu respectively. The sidechain conformations of Ser 21, Asn 41, Asp 136, Ser160, Glu 183 and the main chain conformation ofThr 150 have been changed to get the best fit of themodel to the electron density map. A total of 126waters have been added to the model. At conver-gence the R-factor was 18.2%, geometric rms devia-tions of 0.007A for bond lengths and 1.931° for bondangles. Figure 4 shows the Tyr 12 residue.

The goodness of the geometry of the model hasbeen checked using PROCHECK.48 No residues werein unfavourable regions. The 2Fo-Fc map contouredat the icr level shows generally good connectivity

. . along the main chain, apart from the loop regions ofFigure 4: 2Fo-Fc map at 1.5o level showing the Tyr 12 . . .the protein (especially in the region of residues 116residue from Co concanavalin A to 124 where the map is very poor).

3.4 Filling the low resolution hole in Laue data using fine angular interval datacollection for tetragonal lysozyme

Good quality lysozyme crystals were available to us from experiments that were conducted aboard theSpacehab-1 NASA Space Shuttle mission.49'5° Utilising these crystals and ground controls Laue data were col-

SPIE Vol. 2521 / 167


lected on station 9.5 of the Daresbury Synchrotron Radiation Source.'6 A MAR image plate detector was usedto record 31 images at 2° intervals each with an exposure time of 500 ms from one of the microgravity growncrystals. A ground control crystal was used to collect 23 images at 3° intervals again with 500 ms exposure time.The synchrotron was operating at 2.0 GeV with an average current of 210 mA during the data collection run. Toattenuate the longer wavelengths a 0.4 mm aluminium foil was used in the beam path.

The data were processed using the Daresbury Laue processing suite'3 with a )tmjn of 0.40A, Amaz of 1.55Aand d of 2.50A. Data reduction used the CCP4 programs38 ROTAVATA and AGROVATA.

C

E0C)

Figure 5: Percentage completeness vs d values for Laue data from tetragonal lysozyme crystals collected at station9.5. Data are shown at 2, 4, 10, and 30° angular intervals respectively for an overall angular range of 60°. Lowresolution cutoffs at 50% completeness for protein crystal Laue data have been extracted from the graph toillustrate the interest in low resolution LOT data between dmaz < A < 00 A.

The data were used to refine the structure of lysozyme against coordinates deposited at the PDB, (6LYZ),36'5'as a starting model using the program XPLOR.37 Subsequently electron density maps were calculated using theCCP438 software package and the model fitted to the maps manually using FRODO39 and later O.'°

The effect of the improved completeness at low resolution using the combination of small angular steps andmultiplicity deconvolution is manifest in the electron density maps. Figure 6 shows this clearly by homing in onthe Phe 38 residue. The connectivity of the map at the 2° angular resolution step is good but quality decreases,as would be expected, towards larger angular intervals in a gradual way. The real space R.factors (RSR) areplotted per residue for each Laue data set in figure 7. Hence the ease with which recently available automaticreadout detectors can be utilised is important, as is the extra sensitivity of those devices over photographic filmin realising routine measurement of many more Lane diffraction images per crystal.

168 ISPIE Vol. 2521

To study the effects of data collection at different angular intervals several subsets of the data were examined;the complete data set of 31 images from the microgravity crystal making up 60° in 2° intervals, 16 images makingup 60° in 4° intervals, 6 images making up 60° in 10° intervals and finally 3 images with 30° degree intervals.The data were processed using firstly singlets only and secondly with multiplet deconvolution35 and the resultingcompleteness of both data sets is shown in figure 5. The low resolution cutoffs at 50% completeness are shownproviding evidence of the complementary nature of LOT (section 4) to Lane data collection.

2 degree combined ..-.2 degree singles .+—•

4 degree combined -0--4degree singles -e.-

10 degree combined -10degree singles -'S--

30 degree combined -a--30 degree singles -a--

singlesdmax A

combineddmax A

2°4°10°30°

86816.04.9

93897.85.0

3 4 5 6 7 8 9 10 11d value


Figure 6: 2Fo-Fc Laue electron density maps at the la level showing the Phe 38 residue in tetragonal lysozymefor data collected with (a) 2° angular steps, (b) 4° angular steps, (c) 10° angular steps, and (d) 30° angular steps.Total angular coverage of 60°. For details see section 3.4.

SPIE Vol. 2521 / 169

(a) (b)

(c) (d)


0.50 i i i ' i i i i i i i i0.45 • !\ 2degree

I'

0.150.10 • I 1 1 I 1 I I t I I

10 20 30 40 50 60 70 80 90 100 110 120residue number

Figure 7: The Real-Space R-factors are shown on a per-residue basis, for the 2°, 4°, 10° and 30° angular intervalLane data

4 MONOCHROMATIC LOT DATA IN PROTEINCRYSTALLOGRAPHY

41 Lysozyme LOT crystal test data

LOT patterns (e.g. figure 9a) were collected at SItS Daresbury station 9•5,16 the SRS operated at 2.0 GeVand 274.5 mA with a 5 T Wiggler. An X-ray beam of wavelength 0.92A with a horizontal beam divergence of 1mrad and 0.3 mrad in the vertical, and a 0.2 mm collimator was used. The crystal-to-detector distance 232 mmwas chosen to give a resolution of 2.5A at the outer edges of the detector. Data were collected on a 90 mm radiusMAR Image Plate( 1200 x 1200 pixels and 0.15 mm pixel size). Two data sets were collected, each covering 68°oscillation range in total. The first data set (LOT-i) contains 1 image of 39° and 1 image of 29° . The seconddata set (LOT-2) contains 6 images of 10° and 1 image of 8°. The exposure time was 1 min/°. The images for thetwo LOT data sets were processed using MOSFLM,52 scaled using SCALA and merged usng AGROVATA bothfrom the CCP438 program suite. LOT-2 data (truncated at 4A) were combined with Laue data (2A) using theSCALA program, and then merged into unique data set using AGROVATA. The final R-factors and completenessare summarised in Table 1.

data sets Rmerge(I)

completeness00 tO 2dmin 2dmin to dmjn 00 to dmin

LOT-iLOT-2Lane

Combined

9.2%10.5%8.8%11.4%

88.6%87.5%50.5%94.0%

73.2%94.7%77.9%77.9%

75.4%93.7%74.1%80.1%

Table 1: Rmerge and completeness for the four data sets discussed in section 4. Note the dmjn'S are 2.5A forLOT-i and LOT-2, 2.OA for Laue and Combined (ie.LOT-2 truncated to 4A + Lane to 2A)

The structure refinement of lysozyme was carried out using four data sets, ie LOT-i, LOT-2, Laue dataset (singles and deconvoluted multiples) at 2.OA (Laue), and LOT-2(to 4A) combined with Lane (to 2.OA)(Combined). The starting models used for LOT-2, Lane and Combined were from the PDB (6LYZ),36 whilst

170/SPIE Vol. 2521


0.40

0.35L 0.300C,cth 0.20

0.15

0.10

Figure 8: The Real-Space R-factors are shown on a per-residue basis, for the Combined set, the Laue set, LOT-iand LOT-2

LOT-i used the final protein model from LOT-2. The X-PLOR37 program was used to refine the structure (40cycles of positional refinement followed by 20 cycles of temperature refinement for each run). The 040 prog wasused to examine the 2Fo-Fc and Fo-Fc maps after each run of XPLOR. PEAKMAX and WATERPEAK(CCP438programs) were used to find possible water positions with peaks greater than 3cr in the Fo-Fc maps and goodhydrogen-bond contacts. Some residues were adjusted to fit the 2Fo-Fc maps using O.° Figures 9b, 9c and 9d,show residue Phe 38 for the three data sets.

4.2 Structure refinement results

The crystallographic R-factors and R.M.S deviations of bond lengths, bond angles and torsional angles fromideality are summarised in Table 2. The Real-Space R-factors40'53'54 (RSR) are also calculated which give direct

data setsIR-factors(F)

R.M.S deviationsbond lengths bond angles torsional angles

LOT-iLOT-2Laue

Combined

i7.6%i7.0%19.9%21.0%

o.oiiAo.oioAo.oiiAo.oioA

2.04001.91602.i8i°2.0160

22.803°22.826°22.762°22.475°

Table 2: Refinement R-factors and r.m.s deviations for the four data sets discussed in section 4

insights into the quality of fitting between protein models and the electron density maps on a per-residue basis.They are shown in figure 8 for the four data sets. The average RSR factors for LOT-i, LOT-2, Laue and Combinedare 25.2%, 23.1%, 29.8% and 24.7% respectively.

4.3 LOT development

With the good quality of data from LOT patterns to date, we are now exploring the use of 360° rotations ontoa single image. For a full rotation of a crystal, each reciprocal lattice point will cross the Ewald sphere twice,

SPIE Vol. 2521 I 171

10 20 30 40 50 60 70 80 90 100 110 120residue number


Figure 9: Tetragonal lysozyme LOT data results (a) LOT-i diffraction image recorded on station 9.5 (39° range),(b) 2Fo-Fc electron density from LOT-i data for Phe 38 at 2.5A, (c) 2Fo-Fc electron density from LOT-2 datafor Phe 38 at 2.5A, (d) 2Fo-Fc electron density from combined (ie. LOT-2 truncated at 4Aand Laue data to2A). For details see section 4.

172 / SPIE Vol. 2521

(a) (b)

(c) (d)


therefore, forming two distinct spots on the 2-D detector. We define here four categories of spots possibly existingin a LOT pattern. The percentage of each category depends on the resolution limit, crystal symmetry, crystalorientation and oscillation range. We define these spots as 'single spots', 'coincident symmetric spots', 'clusteringsymmetric spots' and 'overlap spots'. Single spots are spots from a single reflection. Coincident symmetricspots involve symmetry-related reflections(eg. (5,0,6), (-5,0,6), (0,5,6), (O,-5,6) which form a symmetric spot withmultiplicity equal to 4 in a tetragonal space group with the crystal perfectly aligned and cK alongthe rotationaxis) situated exactly at the same position(ie. exact overlaps) but without overlap with any other (h k 1) group.When the crystal is mis.set, some coincident symmetric spots may split into single spots, coincident symmetricspots with lower multiplicities or 'clustering symmetric spots'. Some coincident symmetric spots or single spotsmay overlap with other (h,k,l) groups forming 'overlap spots'. The coincident symmetric spots can be integratedas single spots and the real intensity of each member is the total integrated intensity divided by their multiplicities.The clustering symmetric spots and overlap spots will be difficult to be deconvoluted into individual parts andmay, in the early stages at least, be ignored.

Simulations of 3600 LOT patterns using lysozyme(P43212) as a test sample have been carried out at 5.OAresolution with the crystal perfectly aligned, misset along one axis, and misset along two axes. The statistics ofnumbers of spots and percentage completeness are summarised in Table 3. The very high completeness of data

missettingangles along

a ,b

totalreflections

singlereflections

-

coincidentsymmetricreflections

spatialoverlap

reflections

clusteringsymmetricreflections

uniquereflec-tions

complete-ness

00,0000,601°,6°

158881589215896

0 0%372 2%9388 59%

6520 41%13500 85%164 1%

9368 59%2020 13%4932 31%

0 0%0 0%

1412 9%

260620590

41.7%99.5%94.7%

Table 3: Statistics of 3600 protein crystal (tetragonal lysozyme) LOT patterns for different crystal missettings.Note. The simulations assume the detector type is a 15 cm radius MAR image plate, resolution limit of 5A andspot size 0.7 mm x 0.7 mm.

suggests that 3600 LOT data collection is a feasible way to fill in the low-resolution hole ( oo to dmaz)in theLaue data, and allows the whole data to be recorded in a single image with the highest redundancy possible andavoiding any time overheads of detector read-out time. Most importantly of all if one contemplates rotating acrystal at the angular speed of a rotor arm in an ultracentrifuge (as a guide to a practical limit eg. 60000 rpm),then 1 revolution would be covered in 1 millisecond. At such speeds no detailed sequence of synchronization ofshutters opening or closing is wanted and hence the recording of a full 3600 f data onto one image is thereforebeing explored as a possible rapid diffraction data acquisition method.

5 SUMMARY: X—RAYS, ELECTRONS AND NEUTRONS

The opening up of the dimension of structural changes as a function of time or environmental conditions(temperature, pressure) is of considerable interest.4'29 Since structure is related to function, insights into lifeor materials processes can be obtained. Critically the direct experimental monitoring of the perturbation ofstructures is needed to provide the basis to tether computer graphics simulations. In crystallography"3'4 overfibre diffraction2 the huge challenge is to survey 3-D spatial rather than 2-D spatial X-ray data respectively intime. This has presented challenges in data collection strategies (Laue and monochromatic) as well as sources,optics and detector developments, and choice of suitable structural systems.4'55

In comparing X-rays and electrons, obviously the electron cross section of interaction is much larger andso very thin samples can be studied (e.g. proteins in membranes); this has the advantage that 'direct' freeze

SPIE Vol. 2521 1 173


trapping of altered structures on the millisecond time scale is feasible.56'57 For 3-D samples such as crystals thetime needed to freeze such a sample is exacerbated but of course it is 3-D crystals that provide the greateststructural detail. Cryo-crystallography is an option where processes are slowed down58 but opens up questions ofdirect relevance of such structures to the in vivo state.59 The continued development of rapid data collection forX-ray crystallography is therefore important.

The weakness of X-rays is in locating protons (H atoms) and yet this is a large area of enzymology. Neutronslend themselves to finding protons. White beam synchrotron X-ray techniques" are finding adaptation at thermalneutron sources for quicker neutron data collection.6°

6 ACKNOWLEDGEMENTS

We are grateful to the staffs of the synchrotron radiation sources at Daresbury SRS, Brookhaven NSLS andGrenoble ESRF for the provision of beam. SEtS Station 9.5 has been funded in large part by the Swedish ResearchCouncil (NFR) as well as the UK SERC (now EPSRC) and the UK MRC to whom we are extremely grateful.The ESRF is a collaborative project involving twelve European partners. The EC Science Stimulation Plan andthe SERC (now EPSRC) are thanked for grant support to JRH and NMA. EPSRC provides research studentshipsfor GB, AD, and EHS. The family of Yeu Perng Nieh are thanked for his support. ST is funded by a Universitàdi Theste scholarship. Dr. J. Raftery is also thanked for computational support. The continuing support andinterest of EEV in CCD developments is of immense benefit to this continuing work.

7 REFERENCES1 J. R. Helliwell. Macromolecular Crystallography with Synchrotron Radiation, Cambridge University Press,

UK, 1992.

2 G. Rosenbaum, K. C. Holmes and J. Witz. "Synchrotron radiation as a source for X-ray diffraction," Nature,230:434, 1971.

3 J. R. Helliwell. "Optimisation of anomalous scattering and structural studies of proteins using synchrotronradiation," in Proceedings of the Daresbury Study Weekend, number DL/SCI/R13, pp. 1—6, 1979.

4 D. W. J. Cruickshank, J. R. Helliwell and L. N. Johnson, editors. Time-resolved macromolecular crystallog-raphy, The Royal Society and Oxford University Press, 1992.

5 J. R. Helliwell. "Synchrotron X-ray Crystallography Techniques: Time-Resolved Aspects of Data Collection,"Phil. Trans. R. Soc. Lond., (340):221—232, 1992.

6 K. Moffat, D. M. Szebenyi and D. Bilderback. "X-ray Laue Diffraction from Protein Crystals," Science,223:1423—1425, 1984.

7 5. Weisgerber and 3. R. Helliwell. "High-resolution Crystallographic Studies of Native Concanavalin A usingRapid Laue Data Collection Methods and the Introduction of a Monochromatic Large..angle OscillationTechnique (LOT)," J. Chem. Soc. Faraday Trans., 89(15):2667—2675, 1993.

8 N. Sakabe, S. Ikemizu, K. Sakabe, T. Higashi, A. Nakagawa and N. Watanabe. "Weissenberg camera formacromolecules with imaging plate data-collection system at the Photon Factory - Present status and futureplan," Rev. Sci. Instram., 66(2):1276—1281, 1995.

9 I. A. Andersson, I. J. Clifton, V. Fulop and 3. Hajdu. High speed high resolution data collection on spinachRubisco using a Weissenberg camera at the Photon Factory, in Crystallographic Computing V. Edited byD. Moras et al.. Oxford University Press, 1991.

174 ISPIE Vol. 2521


10 W. L. Bragg. The Development of X-ray Analysis, p. 137, Bell and Sons, 1975.

11 D. W. J. Cruickshank, 3. R. Helliwell and K. Moffat. "Multiplicity Distribution of Reflections in LaueDiffraction," Acta Cryst., (43):656—674, 1987.

12 J. R. Helliwell. "Protein crystallography with synchrotron radiation," J. Molec. Struct., (130):63—91, 1985.

13 J. R. Helliwell, J. Habash, D. W. J. Cruickshank, M. M. Harding, T. Greenhough, J. W. Campbell, I. J.Clifton, M. Elder, P. A. Macun, M. Z. Papiz and S. Zurek. "The Recording and Analysis of SynchrotronX-radiation Laue Diffraction Photographs," J. Appi. Cryst., 22:483—497, 1989.

14 M. Lindahl, A. Liljas, J. Habash, S. Harrop and J. R. Helliwell. "The Sensitivity of the Synchrotron LaueMethod to SmaU Structural Changes: Binding Studies of Human Carbonic Anhydrase II (HCAII)," ActaCr'yst., B(48):281—285, 1992.

15 A. D. Cameron, S. 3. Smerdon, A. J. Wilkinson, J. Habash, J. R. Helliwell, T. Li and J. S. Olson. "DistalPocket Polarity in Ligand Binding to Myoglobin: Deoxy and Carbonmonoxy Forms of a Threonine68 (Eli)Mutant Investigated by X-ray Crystallography and Infrared Spectroscopy," Biochemistry, 32:13061—13070,1993.

16 R. C. Brammer, J. R. Helliwell, W. Lamb, A. Liljas, P. R. Moore, A. W. Thompson and K. Rathbone. "Anew protein crystallography station on the SRS wiggler beamline for very rapid Laue and rapidly tunablemonochromatic experiments .1. Design principles, ray tracing and heat calculations," Nucl. Instrum. MethodsA, 27l(3):678—687, 1988.

17 N. M. Allinson, P. D. Carr, M. Colapietro, M. M. Harding, J. R. Helliwell, A. W. Thompson and S. Weis-gerber. "Time-Resolved Synchrotron Laue Diffraction and its Application in Structural Molecular Biologyand Materials Science," Phase Transitions, 39:145—160, 1992.

18 A. Deacon, J. Habash, S. J. Harrop, J. R. Helliweil, W. N. Hunter, G. A. Leonard, M. Peterson, A. Hadener,A. J. Kaib (Gilboa), N. M. Allinson, C. Castelli, K. Moon, S. McSweeney, A. Gonzalez, A. W. Thompson,S. EaliCk, D. M. Szebenyi and R. M. Walter. "SR Instrumentation for optimized anomalous scattering andhigh resolution structure studies of proteins and nucleic acids," Rev. Sci. Instrum., 2(66):l287—l292, 1995.

19 J. R. Helliwell, T. J. Greenhough, P. D. Carr, S. A. Rule, P. R. Moore, A. W. Thompson and J. S. Wor-gan. "Central data-collection facility for protein crystallography, small-angle diffraction and scattering atthe Daresbury-laboratory synchrotron rariation source (SRS), England," Journal of Physics E-ScientificInstraments, 15(l2):1363—1372, 1982.

20 J. R. Helliwell, M. Z. Papiz, I. D. Glover, A. W. Habash, J. Thompson, P. R. Moore, N. Harris, D. Croftand E. Pantos. "The Wiggler Protein Crystallography Workstation at the Daresbury SRS. Progress andResults," Nucl. Instr. and Meth., (246):617—623, 1986.

21 P. F. Lindley, E. M. H. Duke and C. Nave et al.to be published.

22 Y. P. Nieh and 3. R. Helliwell. "Time differences between Friedel reflections: Accuracy of crystal setting andrequirements on beam stability," J. Synchrotron Rad., 2:79—82, 1995.

23 I. 3. Clifton, M. Elder and 3. Hajdu. "Experimental Strategies in Laue Crystallography," J. Appl. Cryst.,24:267—277, 1991.

24 3. R. Helliwell. "Macromolecular crystallography beamline," in ESRF Red Book, pp. 329—340, 1987.

25 M. Wulif ESRF Newsletter, 1994.

26 3. P. Moy and S. Gibney. "Low Energy X-ray Image Intensifier Development Phase 1 Report," PreliminaryReport EXP/JGM/SG/92/91, ESRF, Grenoble, France, December 1992.

SPIE Vol. 2521 / 175


27 N. M. Allinson, R. Brammer, J. R. Helliwell, S. Harrop, B. G. Magorrian and T. Wan. "Charge coupleddevice (CCD) area detector for on-line (40-80 ms) acqusition of Laue diffraction data from protein crystals,"tJ. X-ray Sci. Technol., 1(143), 1989.

28 N. M. Allinson, M. Colapietro, 3. R. Helliweil, K. J. Moon, A. W. Thompson and S. Weisgerber. "Charge-Coupled Irnagers for Time-Resolved Macromolecular Crystallography," Rev. Sci. Instrum., 63(1):664—666,Jan 1992.

29 P. Coppens. Synchrotron Radiation in Crystallography, Academic Press, 1992.

30 K. Moon, N. M. Allinson and J. R. Helliwell. "High-Speed CCD Aquisition System," Nuci. Instr. and Meth.,A(348):631—634, 1994.

31 J. Hajdu, S. C. Almo, G. K. Farber, J. K. Prater, G. A. Petsko, S. Wakatsuki, I. J. Clifton and V. Fulöp.Edited by D. Moras et al.. "On the Limitations of the Laue Method when Applied to Crystals of Macro-molecules," Crystallographic Computing V, pp. 29—49. OUP, 1991.

32 Z. Ren and K. Moffat. "Laue crystallography for analysing rapid reactions," J. Synchrotron Rad., (1):18—82,1994.

33 A. Cassetta, A. Deacon, C. Emmerich, J. Habash, 3. R. Helliweil, S. McSweeney, E. Snell, A. W. Thompsonand S. Weisgerber. "The Emergence of the Synchrotron Laue Method for Rapid Data Collection from ProteinCrystals," Proc. R. Soc. Lond., A(442):177—192, 1993.

34 A. Deacon, J. Habash, A. W. Thompson, 3. R. Helliwell, S. McSweeney and E. H. Snell. "The developmentof a transmission mirror to control Amaz in Laue diffraction," J. Synchrotron Rad., 1995, To be submitted.

35 J. w. Campbell and Q. Hao. "Evaluation of Reflection Intensities for the Components of Multiple LaueDiffraction Spots. II. Using the Wavelength Normalisation Curve," Acta Cryst., A(49):889—893, 1993.

36 R. Diamond. "Real space refinement of the structure of hen egg-white lysozyme," J. Mol. Biol., (82):371—391,1974.

37 A. T. Brünger. X-PLOR Version 3.1 A System for X-ray Crjistallography and NMR, Yale University Press,New Haven and London, 1992.

38 Collaborative Computational Project Number 4. "The CCP4 Suite: Programs for Protein Crystallography,"Acta Cryst., D(50):760—763, 1994.

39 T. A. Jones and L. Liljas. "Crystallographic refinement of macromolecules having non-crystallographicsymmetry," Acta Cryst., (40):50—57, 1984.

40 T. A. Jones, J. Y. Zou, S. W. Cowan and M. Kjelgaard. "Improved methods for building protein models inelectron density maps and the location of errors in these models," Acta Cr'yst., A(47):11O-.119, 1991.

41 C. Rarnakrishnan and G. N. Ramachandran. "Stereochemical criteria for polypeptide and protein chainconformation," Biophys. J., (5):909—933, 1965.

42 J. W. Campbell, A. Deacon, J. Habash, J. R. Helliwell, S. McSweeney, H. Quan, J. Raftery and E. Sneil."Electron Density Maps of Lysozyme Calculated using Synchrotron Laue Data Comprising Singles andDeconvoluted Multiples," Bull. Mater. Sci., 17(1):1—18, 1994.

43 T. L. Blundell and L. N. Johnson. Protein Crystallography, Academic Press, 1976.

44 P. J. Artymiuk, C. C. F. Blake, D. W. Rice and K. S. Wilson. "The structures of the monoclinic andorthorhombic forms of hen egg-white lysozyme at 6A resolution," Acta Cr'yst., B(38):778—783, 1982.

45 J. W. Campbell. "LAUEGEN, an X-windows-based program for the processing of Laue X-ray diffractiondata," J. Appi. Cryst., 28:228—236, 1995.

176 ISPIE Vol. 2521


46 P. D. M. Fitzgerald. "MERLOT, an integrated package of computer programs for the determination ofcrystal structures by molecular replacement," Acta Cryst., (21):273—278, 1988.

47 J. H. Naismith, J. Habash, S. Harrop, J. R. Helliwell, W. N. Hunter, T. C. M. Wan and S. Weisgerber."Refined structure of Cadmium-Substituted concanavalin A at 2.OA resolution," Acta Cnjst., (50):561—571,1993.

48 R. A. Laskowski, M. W. MacArthur, D. S. Moss and 3. Thornton. "PROCHECK -A program to check thestereochemical quality of protein structures," J. Appi. Cryst., 26(2):283—291, 1993.

49 S. Weisgerber and 3. R. Helliwell. "Chicken Egg-White Lysozyme Crystals Grown in Microgravity and onEarth for a Comparison of their Respective Perfection," Joint CCP4 and ESF-EA CBM Newsletter on ProteinCrystallography, (29):1O—13, 1993.

50 3. R. Helliwell, E. H. Snell and S. Weisgerber. "An Investigation of the Perfection of Lysozyme Protein Crys-tals Grown in Microgravity (Spacehab-1 and IML-2) and on Earth," in Proceedings of the Berlin MicrogravityMeeting.. Springer Verlag, 1995, in press.

51 C. C. F. Blake, G. A. Mair, A. C. T. North, D. C. Philips and V. R. Sarma Proc. Roy. Soc. (London),B(167):365, 1967.

52 A. J. Wonacott, S. Dockeril and P. BriCk MOSFLM program. Unpublished notes.

53 R. Diamond. "A Real-Space Refinement Procedure for Proteins," Acta Cryst., A(27):436—452, 1971.

54 R. K. Wierenga, K. H. Kalk and W. G. 3. Hol. "Structure determination of the glycosomal triosephosphateisomerase from trypanosoma-brucei-brucei at 2.4A resolution," J. Mol. Biol., (1):109—121, 1987.

55 E. F. Pai. "Time-resolved macromolecular crystallography," Current Opinion in Structural Biology, (2):821—827, 1992.

56 S. Subramaniam, M. Gerstein, D. Desterfelt and R. Henderson. "Electron diffraction analysis of structuralchanges in the photocycle of bacteriororhodopsin," EMBO J., 12:1—8, 1993.

57 J. Berriman and N. Unwin. "Analysis of transient structure by cryo-microscopy combined with rapid mea-suring of spray droplets," Ultramicroscopy, 1995, In press.

58 T. Y. Teng, V. Srajer and K. Moffat. "Photolysis-induced structural-changes in single-cyystals of carbon-monoxy myoglobin at 40 K," Nature Structural Biology, 1(10):701—705, 1994.

59 B. F. Rasmussen, A. M. Stock, D. Ringe and G. A. Petsko. "Crystalline ribonuclease-A loses function belowthe dynamic transition at 220-K," Nature, 357(6377):423—424, 1992.

60 C. Wilkinson and M. S. Lehmann. "Quasi-Laue neutron diffractometer," Nucl. Inst. Methods, 310(1-2):411—415, 1991.

SPIE Vol. 2521 / 177


time-resolved biological and perturbation chemical crystallography… · 2019-11-27 ·...

Documents